Legal claims defining the scope of protection, as filed with the USPTO.
1. A computer-implemented method for processing a plurality of recorded calls, the method comprising: defining a set of one or more categories having a corresponding set of conditions that associate the recorded calls with the categories, wherein the recorded calls are categorized in terms of their root causes; in a graphical user interface, accepting a sub-categorization request sent from an analyst, requesting to divide a category from among the categories into lower-level categories; responsively to the sub-categorization request, processing the recorded calls that are associated with the category using a clustering process that divides the recorded calls into a plurality of clusters so as to automatically suggest the lower-level categories, wherein each of the lower-level categories is from a different one of the plurality of clusters; presenting the automatically-suggested lower-level categories to the analyst and accepting direction from the analyst with respect to the automatically-suggested lower-level categories; responsively to the direction, constructing a hierarchical structure representing the categories by dividing the category into the lower-level categories; presenting output based on the hierarchical structure to the analyst; and analyzing the recorded calls using the hierarchical structures.
2. The method according to claim 1 , wherein the categories in the hierarchical structure describe respective types of root causes that caused the recorded calls to be included in the plurality.
3. The method according to claim 1 , and comprising performing multiple iterations of accepting sub-categorization requests from the analyst, automatically suggesting respective sets of the lower level categories responsively to the sub-categorization requests, presenting the automatically-suggested sets of the lower-level categories to the analyst, and iteratively updating the hierarchical structure of the categories responsively to accepting analyst directions.
4. The method according to claim 1 , wherein processing the recorded calls comprises identifying textual terms in the recorded calls and determining similarities among the recorded calls based on the identified textual terms.
5. The method according to claim 1 , wherein the recorded calls comprise textual terms having respective confidence scores, and wherein processing the recorded calls comprises determining similarities among the recorded calls based on the textual terms and the confidence scores.
6. The method according to claim 1 , wherein the recorded calls comprise textual terms, and wherein the conditions comprise a condition that operates on at least the textual terms of a tested data item, such that the tested data item is associated with one of the categories when the condition is fulfilled.
7. The method according to claim 1 , wherein the recorded calls comprise textual terms, and wherein the conditions comprise a condition that operates on at least the textual terms of a tested data item and produces a numerical association metric quantifying an association of the tested data item with one of the categories.
8. The method according to claim 1 , wherein dividing the category into the lower-level categories comprises defining respective conditions for associating subsequent recorded calls with the lower-level categories.
9. The method according to claim 1 , wherein processing the recorded calls comprises automatically generating respective suggested titles for the lower-level categories.
10. The method according to claim 1 , wherein the recorded calls comprise textual terms, wherein defining the categories comprises defining respective subsets of the textual terms that characterize the categories, and comprising accepting from the analyst an expansion request requesting to expand a subset of the textual terms that characterizes a respective target category and, responsively to the expansion request, automatically determining one or more additional textual terms that characterize the target category and adding the additional textual terms to the subset.
11. The method according to claim 1 , wherein the recorded calls comprise recorded speech.
12. The method according to claim 1 , wherein the recorded calls comprise one of recorded sessions conducted with customers, recorded communication sessions intercepted from a communication network and textual documents.
13. Apparatus for processing a plurality of recorded calls, the apparatus comprising: an interface, which is operative to interact with analyst; and a processor, which is arranged to accept a definition of a set of one or more categories having a corresponding set of conditions that associate the recorded calls with the categories, wherein the recorded calls are calls, and the calls are to be categorized in terms of their root causes, to accept from the analyst via the interface a sub-categorization request, requesting to divide a category from among the categories into lower-level categories, to process the recorded calls that are associated with the category responsively to the sub-categorization request using a clustering process that divides the recorded calls into a plurality of clusters so as to automatically suggest the lower-level categories, to present the automatically-suggested lower-level categories to the analyst, to accept direction from the analyst with respect to the automatically-suggested lower-level categories, to construct, responsively to the direction, a hierarchical structure representing the categories by dividing the category into the lower-level categories, and to present output based on the hierarchical structure to the analyst via the interface, and to analyze categories of recorded calls using the hierarchical structures; wherein each of the lower-level categories is from a different one of the plurality of clusters.
14. The apparatus according to claim 13 , wherein the categories in the hierarchical structure describe respective types of root causes that caused the recorded calls to be included in the plurality.
15. The apparatus according to claim 13 , wherein the processor is arranged to perform multiple iterations of accepting sub-categorization requests from the analyst, automatically suggesting respective sets of the lower-level categories responsively to the sub-categorization requests, presenting the automatically-suggested sets of the lower-level categories to the analyst, and iteratively updating the hierarchical structure of the categories responsively to accepting analyst directions.
16. The apparatus according to claim 13 , wherein the processor is arranged to identify textual terms in the recorded calls and to suggest the lower-level categories by determining similarities among the recorded calls based on the identified textual terms.
17. The apparatus according to claim 13 , wherein the recorded calls comprise textual terms having respective confidence scores, and wherein the processor is arranged to suggest the lower-level categories by determining similarities among the recorded calls based on the textual terms and the confidence scores.
18. The apparatus according to claim 13 , wherein the recorded calls comprise textual terms, and wherein the conditions comprise at least one condition, which operates on at least the textual terms of a tested data item and is selected from a group of condition types consisting of a Boolean condition that associates the tested data item with one of the categories when the Boolean condition is fulfilled, and a soft condition that produces a numerical association metric quantifying an association of the tested data item with the one of the categories.
19. The apparatus according to claim 13 , wherein the processor is arranged to define respective conditions for associating subsequent recorded calls with the lower-level categories.
20. The apparatus according to claim 13 , wherein the recorded calls comprise recorded speech.
Unknown
April 21, 2015
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.